The Basics of Graphical Models

David M. Blei
Columbia University

October 3, 2015

Introduction

‘ These notes follow Chapter 2 of An Introduction to Probabilistic Graphical Models by

Michael Jordan. Many figures are taken from this chapter.

‘ Consider a set of random variables fX1 ; : : : ; Xn g. We are interested in

 Which variables are independent?
 Which variables are conditionally independent given others?
 What are the marginal distributions of subsets of variables?
These questions are answered with the joint distribution P .X1 ; : : : ; Xn /.
 Marginalization is answered by summing over the joint.
 Independence is answered by checking factorizations.
‘ Assume the variables are discrete (i.e., categorical). The joint distribution is a table
p.x1 ; : : : ; xn /. Each cell is non-negative and the table sums to one. If there are r possible
values for each variable then the naïve representation of this table contains r n elements.
When n is large, this is expensive to store and to use.

Graphical models provide a more economic representation of the joint distribution by

taking advantage of local relationships between random variables.

Directed graphical models

‘ A directed graphical model is a directed acyclic graph. The vertices are random variables
X1 ; : : : ; Xn ; edges denote the “parent of” relationship, where i are the parents of Xi .

Here is an example:

1
 X4
X2
X6
X1

X3 X5

The random variables are fX1 ; : : : ; X6 g, e.g., 6 D fX5 ; X2 g.

‘ The graph defines a factorization of the joint distribution in terms of the conditional
distributions [Link] j xi /.
p.x1W6 / , p.x1 /p.x2 j x1 /p.x3 j x1 /p.x4 j x2 /p.x5 j x3 /p.x6 j x2 ; x5 /

In general,
n
Y
p.x1Wn / , [Link] j xi /: (1)
i D1

(Note that we can use a set in the subscript.) This joint is defined in terms of local probability
tables. Each table contains the conditional probabilities of a variable for each value of the
conditioning set.

‘ By filling in the specific values for the conditional distributions, we produce a specific
joint distribution of the ensemble of random variables. Holding the graph fixed, we can
change the local probability tables to obtain a different joint.

Now consider all possible local probability tables. We see that the graphical model represents
a family of distributions. The family is defined by those whose joint can be written in terms
of the factorization implied by the graph. It is important to notice that this is not all
distributions over the collection of random variables.

Graphical models represent a family of distributions.

‘ What is the advantage of limiting the family? Suppose x1Wn are binary randomPvariables.
The full joint requires 2n values, one per entry. The graphical model joint requires niD1 2ji j
entries. We have replaced exponential growth in n by exponential growth in ji j.

In statistical and machine learning applications, we represent data as random variables and
analyze data via their joint distribution. We enjoy big savings when each data point only
depends on a couple of parents.

2
‘ This is only part of the story. In addition to economic representation:
 Graphical models give us inferential machinery for computing probabilistic quantities
and answering questions about the joint, i.e., the graph. The graph determines, and thus
lets us control, the cost of computation. (And, as an aside, these same considerations
apply when thinking about data and statistical efficiency. But this is less looked at in
the graphical models literature.)
 Finally, graphs are more generic than specific joint distributions. A graphical model
of binary data can be treated with similar algorithms as a graphical model with r-ary
data. And, later, we will see how the same algorithms can treat discrete / categorical
variables similarly to continuous variables, two domains that were largely considered
separately. For example, graphical models connect the algorithms used to work with
hidden Markov models to those used to work with the Kalman filter.

Directed graphical models and conditional independence

‘ Recall the definition of independence

xA ?
? xB ! [Link] ; xB / D [Link] /[Link] / (2)
! [Link] j xB / D [Link] / (3)
! [Link] j xA / D [Link] / (4)

And recall the equivalent definitions of conditional independence

xA ?
? xB j xC ! [Link] ; xB j xC / D [Link] ; xB j xC / (5)
! [Link] j xB ; xC / D [Link] j xC / (6)
! [Link] j xA ; xC / D [Link] j xC / (7)

These are questions about factorizations of marginal distributions. They can be answered by
examining—or computing about—the graph.

‘ Recall the chain rule of probability

n
Y
p.x1Wn / D [Link] j x1 ; : : : ; xi 1/ (8)
i D1

In our example

p.x1W6 / D p.x1 /p.x2 j x1 /p.x3 j x1 ; x2 /


p.x6 j x1W5 /: (9)

The joint distribution defined by the graph is suggestive that, e.g.,

p.x4 j x1 ; x2 ; x3 / D p.x4 j x2 /; (10)

3
which means that

x4 ?
? xf1;3g j x2 : (11)

This statement is true. It is one of the basic conditional independence statements.

Let’s prove it:

 Write the conditional as a ratio of marginals

p.x1 ; x2 ; x3 ; x4 /
p.x4 j x1 ; x2 ; x3 / D (12)
p.x1 ; x2 ; x3 /
 Numerator: take the joint and marginalize out x5 and x6
 Denominator: Further marginalize out x4 from the result of the previous step.
 Finally, divide to show that this equals p.x4 j x2 /.
‘ More generally, let I be a topological ordering of the random variables, which ensures
that i occurs in the ordering before i. Let i be the set of indices that appear before i, not
including i . The set of basic conditional independence statements is

fxi ?
? xi j xi g (13)

In our example, one valid topological ordering is I D f1; 2; 3; 4; 5; 6g. This implies the
following independencies,

X1 ? ? 0j0 (14)
X2 ? ? 0 j X1 (15)
X3 ? ? X2 j X1 (16)
X4 ?
? fX1 ; X3 g j X2 (17)
X5 ?
? fX1 ; X2 ; X4 g j X3 (18)
X6 ?
? fX1 ; X3 ; X4 g j fX2 ; X5 g (19)

‘ This is a little inelegant because it depends on an ordering. Here is a graph-specific

definition of the basic conditional independencies. Redefine i to be all the ancestors
of i excluding the parents. Using this definition, we can define the basic conditional
independencies as in Equation (13).

Note: This does not give us all of the possible basic conditional independencies, i.e., all of
those possible by traversing all topological orderings, but that doesn’t really matter.) The
point is this. We can read off conditional independencies by looking at the graph.

‘ We emphasize that these independencies hold regardless of the specific local probability
tables. This gives us an insight about the nature of graphical models:

4
By using the cheaper factorized representation of the joint, we are making certain
independence assumptions about our random variables.

This makes more precise—a little, anyway—the difference between the family specified by
the graphical model and the family of all joints.

Q: Are these the only independence assumptions we are making?

The Bayes ball algorithm and d -separation

‘ Note that a node’s parents separate it from its ancestors. It appears that conditional
independencies are related to the graph and, in particular, to graph separation. We will next
uncover the relationship between graph separation and conditional independence.

To do this, and deepen our understanding of independence and graphical models, we look at
three simple graphs. We ask which independencies hold in these graphs and consider the
relationship to classical graph separation.

‘ The first is a little sequence,

X Y Z

p.x; y; z/ D p.x/p.y j x/p.z j y/: (20)

Here,

X?
? Z j Y: (21)

To see this,
p.x; y; z/
p.x j y; z/ D (22)
p.y; z/
p.x/p.y j x/p.z j y/
D (23)
p.z j y/ x 0 p.x 0 /p.y j x 0 /
P

p.x; y/
D (24)
p.y/
D p.x j y/ (25)

We assert that no other independencies hold. (E.g., is it not true that X ?

? Z.)

5
Here we do not need to do this algebra. This is one of the basic conditional independencies.
The parent of Z is Y , and X is a non-parent ancestor of Z.

Important subtlety: This means that other independencies do not necessarily hold. For some
settings of p.y j x/ it may be true that X ? ? Z. But, not for all. In other words, a more
restrictive family of joints will be contained in the less restrictive family.

In this graph, conditional independence can be interpreted as graph separation. In graphical

models notation, we shade the node that we are conditioning on:

X Y Z .

We can see that Y separates X and Z.

The intuition: X is the “past”, Y is the “present”, Z is the “future”. Given the present, the
past is independent of the future. This is the Markov assumption. This graph is a three step
Markov chain.

‘ The second graph is a little tree,

X Z

p.x; y; z/ D p.y/p.x j y/p.z j y/ (26)

Here we have again that X ?

? Z j Y . We calculate that the conditional joint factorizes,

p.y/p.x j y/p.z j y/
p.x; z j y/ D (27)
p.y/
D p.x j y/p.z j y/ (28)

We assert that no other conditional independencies hold. Again, simple graph separation
indicates independence,

6
 Y

X Z

The intuition behind this graph comes from a latent variable model. In our previous lecture,
this graph describes the unknown coin flipped twice.

As another example, let X be “shoe size” and Z be “amount of gray hair”. In general, these
are dependent variables. But suppose Y is “age”. Conditioned on Y , X and Z become
independent. Graphically, we can see this. It is through “age” that “shoe size” and “gray
hair” depend on each other.

‘ The last simple graph is an “inverse tree”

X Z

p.x; y; z/ D p.x/p.z/p.y j x; z/ (29)

Here the only independence statement is X ?

? Z. In particular, it is not necessarily true that
X?? Z jY.

For intuition, think of a causal model: Y is “I’m late for lunch”; X is “I’m abducted by
aliens”, a possible cause of being late; Z is “My watch is broken”, another possible cause.
Marginally, being abducted and breaking my watch are independent. But conditioned on my
lateness, knowing about one tells us about the likelihood of the other. (E.g., if I’m late and
you know that my watch is broken, then this decreases the chance that I was abducted.)

Alas this independency does not correspond to graph separation.

‘ With these simple graphs in hand, we can now discuss d -separation, a notion of graph
separability that lets us determine the validity of any conditional independence statement in
a directed graphical model.

Suppose we are testing a conditional independence statement,

XA ?
? XB j XC : (30)

7
We shade the nodes being conditioned on. We then decide, using the “Bayes ball” algorithm,
whether the conditioned nodes d -separate the nodes on either side of the independence
relation.

The Bayes ball algorithm is a reachability algorithm. We start balls off at one of the sets of
variables. If they can reach one of the other set then the conditional independence statement
is false.

The balls bounce around the graph according to rules based on the three simple graphs. We
consider a ball starting at X and going through Y on its way to Z. (To be clear, if the move
is allowed, then the next step is for the ball to be at Y and we ask if it can go through Z en
route to another node.)

Note that it does not matter if the source node X and destination node Z are shaded.

‘ Here are the rules:

In addition, there are rules derived by contemplating a ball going through a node and then
back to the source node:

‘ Some examples:
1. Look at our example graph.

8
 (a) X1 ?
? X6 j fX2 ; X3 g? Yes.
(b) X2 ?
? X3 j fX1 ; X6 g? No.
2. A Markov chain is the simple sequence graph with any length sequence. The basic
conditional independencies are that

Xi C1 ?
? X1W.i 1/ j Xi : (31)

But Bayes ball tells us more, e.g.

X1 ?
? X5 j X4
X1 ?
? X5 j X2
X1 ?
? X5 j X2 ; X4

3. Now consider a hidden Markov model which is used, for example, in speech recogni-
tion. The Bayes ball algorithm reveals that there are no conditional independencies
among the observations.
4. (Optional) Look at a tree model, e.g., of genetic sequences in a family tree. What
kinds of independencies do we see?
5. Look at a Bayesian hierarchical regression model. (E.g., consider testing in different
schools.) How are the groups related? What if we know the prior?
‘ Remarks on Bayes ball:
 It’s not an algorithm that is necessarily very interesting to implement. But it’s very
useful to look at graphs—i.e., at structured joint distributions—and understand the
complete set of conditional independence and independence assumptions that are
being made. As we have shown, this is not obvious either from the joint distribution
or the structure alone.
 The idea of a ball bouncing around is a theme that we will come back to. It won’t be
balls, but be “messages” (i.e., information). Just as balls bouncing around the graph
help us understand independence, messages traveling on the graph will help us make
probabilistic computations.

The Hammersley-Clifford theorem

‘ Punchline:

Consider two families of joint probability distributions, both obtained from the graphical
model G.
1. Family of joints found by ranging over all conditional probability tables associated
with G.
2. All joints that respect all conditional independence statements, implied by G and
d -separation.

9
The Hammersley-Clifford theorem says that these families are the same.

‘ More verbose:

We find the first family by varying the local conditional probability tables and computing
the resulting joint from its factorization. This is what we meant earlier when we said that a
graphical model defines a family of probability distributions.

We obtain the second family as follows. First, compute every conditional independence
statement that is implied by the graph. (Use Bayes ball.) Then, consider every joint
distribution of the same set of variables. Note this does not reference the local conditional
probability tables. For each joint, check whether all the conditional independence statements
hold. If one does not, throw the joint away. Those that remain are the second family.

The Hammersley-Clifford theorem says that these two families of joints—one obtained by
checking conditional independencies and the other obtained by varying local probability
tables—are the same.

As stated in the chapter, this theorem is at the core of the graphical models formalism. It
makes precise and clear what limitations (or assumptions) we place on the family of joint
distributions when we specify a graphical model.

Undirected graphical models

In this class, we will mainly focus on directed graphical models. However, undirected graph-
ical models, which are also known as Markov random fields, are a very useful formalism
as well. They are important to learn about to be fluent in graphical models, will be useful
later when we talk about exact inference, and further refine the picture of the relationship
between graphs and probability models.

A definition, via conditional independencies

When discussing directed models, we began with a definition of how to map a graph to a
joint and then showed how the resulting conditional independencies can be seen from the
graph. Here we will go in reverse. Based on an undirected graph, we will first define the
conditional independencies that we want to hold in its corresponding joint distribution. We
will then define the form of that distribution.

Consider an undirected graph G D .V; E/ and three sets of nodes A, B, and C . We will
want a joint distribution such that XA ?
? XC j XB if XB separates XA and XC , in the usual
graph-theoretic sense of separate.

Formally, quoting from the book, “if every path from a node in X_A to a node in X_C
includes at least one node in X_B then we assert that XA ?
? XC j XB .”

10
Again we emphasize that we are representing a family of distributions. These are the
conditional independence statements that (we assert) have to hold. For various instantiations
of the graphical model (i.e., various members of the family) other conditional independencies
may also hold.

Undirected and directed graphical models are different

Consider the families of families expressable by directed and undirected graphical models,
respectively. Not all directed graphical models can be written as an undirected graphical
model, and vice versa.

First consider this directed graph,

X Z

As we said, the only conditional independence statement that is true for this graph is
X? ? Z. We cannot write an undirected graphical model such that this is the only conditional
independence statement, i.e. where X 6?
? Z jY.

Now consider this undirected graph,

[Four nodes connected in a square]

This graph expresses the family characterized by the following independencies,

X??Y j fW; Zg (32)

W ?
?Z j fX; Y g (33)

And these are the only ones.

We cannot write down a directed graphical model such that these are the only two conditional
independence statements. Exercise: Confirm this.

Note that there are types of directed and undirected graphical models that can be written as
either. We will one such important class when we talk about inference. But, in general, we
have just demonstrated that they have different expressive power.

The joint distribution in an undirected graphical model

From the conditional independencies, we will now develop a representation of the joint
distribution. Our goal is to represent the joint as a product of “local” functions, which we

11
will call potential functions, whose arguments are subsets of the random variables,
1 Y
p.x1 ; : : : ; xn / D .xS / (34)
Z
S2S

Here S is a set of nodes, .
/ are arbitrary different potential functions (notation overloaded),
and S is a collection of subsets. (We specify them later.) We would like these functions to
be non-negative but otherwise arbitrary, so we will be satisfied with specifying them up to
a scaling constant Z. (We use this notation to be consistent with the literature; Z is not a
random variable.) This is called the normalizing constant, and will be an important quantity
for much of this course.

We need to define what we mean by “local.” This amounts to choosing the arguments of
each of the potential functions, i.e., choosing the subsets S. Let’s return to the conditional
independencies that we are hoping to assume with this representation. These imply that if
two nodes X1 and X3 are separated by a third X2 then X1 ? ? X3 j X2 . This implies that the
conditional distribution factorizes,

p.x1 ; x3 j x2 / D p.x1 j x2 /p.x3 j x2 /: (35)

This further implies that the three nodes cannot participate in a single potential. Why? If
there were an arbitrary potential function .x1 ; x2 ; x3 / in the joint distribution of Equa-
tion (34) then it would be impossible for the conditional (which, recall, is proportional to
the joint) to factorize across x1 and x3 .

Maximal cliques. This is suggestive that the potential functions should only be defined on
cliques, which are sets of nodes that are fully connected, or subsets of cliques. Because of
the direct connections between all nodes in a clique we are guaranteed that the graph does
not imply any conditional independencies between them; thus it is safe to include them in
the arguments to a potential. Conversely, as we argued above, if two nodes are not directly
connected in the graph then there is a conditional independence statement that we can make.
Thus, they should not appear together in a potential.

In the theory around undirected graphical models, the joint is defined on the set of maximal
cliques, i.e., completely connected components of the graph that cannot be expanded without
breaking complete connectedness. Every node is part of a maximal clique. Thus, we can
write the joint as
1 Y
p.x/ D .xC /: (36)
Z
C 2C

Here, C is the set of maximal cliques and C is a particular clique (i.e., set of nodes). The
normalizing constant is
XY
ZD .xC /: (37)
x C 2C

It is difficult to compute. (Why?) We’ll come back to that later in the semester.

12
This joint distribution respects the set of conditional independence statements implied by
usual graph separability on the underlying graph.

Finally, in practice we often define undirected graphical models in terms of other cliques, in
addition to or instead of maximal cliques. As long as we don’t steer beyond a maximal clique,
this preserves the relationship between graph separation and conditional independence.

Interpreting potentials. The potential functions we set up are arbitrary positive valued
functions. They are not conditional probabilities (necessarily) as in the directed graphical
models case. However, they can be interpreted as providing “agreement” to configurations
of variables that have high probability. If the potential on .x1 ; x2 / is high then the
configuration with those values has higher probability. Though we will not discuss it in
depth, this is how undirected graphical models play a large role in statistical physics (the
field in which they were invented).

Hammersley-Clifford for undirected graphical models. We can state a similar theorem

for undirected graphical models as we did for directed graphical models.

Fix an undirected graph G.

Define one family of distributions by ranging over all possible potential functions over the
maximal cliques of the graph, and calculating the joint distribution in Equation (36).

Define a second family of distributions by looking at all joint distributions over the set of
nodes in the graph and filtering out only those for which the set of conditional independence
statements—defined by graph separability—holds.

These two sets of distributions are the same.

13